recycle energy from lost particles?

[happyjack27]

i've noticed in my sims at least that when ions escape they do so out the cusps at very high and very specific momentum. i recalled the "direct electric energy conversion" mechanism for harnessing the MeV from pb11 fuel. could not the same mechanism be used to gather the energy from these losses and recycle it back into the machine?

[93143]

Probably. I think that might actually be part of the plan... with a positive magrid they gain a standard amount of energy when escaping, so the collectors don't have to worry about a low-energy tail...

[hanelyp]

You need to be careful that direct conversion traps designed for escaping ions don't suck in electrons. A grid in front to hide the collectors potential from electrons should work.

[D Tibbets]

High energy fuel ions escaping has been one of the bones of contention about the potential well depth necessary to contain them sufficiently. This value can be fairly high if I understood A. Carlson's arguments (over a million volts, I don't know if he was considering D-D or P-B11 fusion). But, this ignores the claimed annealing process that Bussard described. If your simulation does not incorporate this process, it is lacking a major component of the claimed Polywell physics.

In any case, if there is a collection grid in place for charged fusion product collection, then it would also reclaim energy from these upscattered fuel ions. They will tend to be perhaps 1-10% of the energy of the fusion ions, so attention would be needed to prevent these upscattered fuel ions from neutralizing and accumulating outside the magrid. It depends in large part on the numbers of these ions compared to the fusion produced ions.

Since the collection grid is outside the magrid, Gauss law would help. I think so long as these grids are placed far enough outside of the magrid, problems with the cusps should be minimal (much like considerations involving the electron guns).

Dan Tibbets

[chrismb]

But, this ignores the claimed annealing process that Bussard described. If your simulation does not incorporate this process, it is lacking a major component of the claimed Polywell physics.

I would suggest that any correct simulation may well lack major components of the claimed Polywell!

Once 'annealing' is observed and measured experimentally, then it would be a fair point. At the moment, as things stand, a simulation that doesn't work as predicted is telling us something.

I still do not see how 'annealing' can actually work, as it is pushing particles into a lower entropy and there is no mechanism doing work on those particles in the edge to avoid thermalisation so I don't see any inference for it. Hand-waving can't 'anneal' particles, so what does when all there appears to be is hand-waving?

[happyjack27]

it loooks from the sims that these plates could be very small and be placed very far from the magrid and still catch most of the ions. esp. if the magrid is charged.

fusion products presumably wouldn't have such a large preference for the cusps (or am i wrong here?) so you can use an altogether separate set of plates for those.

[happyjack27]

i've been giving annealing some thought. and it seems to me that as colder ions away from the center, they adjust to the electric potential field they create slowly and in that sense more finely. thus their total electric potential energy is going to lower a little better and as they do that their movement will be more uniform, as non-uniformity would be pushed out by the e-field. sort of like a flock of geese. so then the question is where does this energy go? my guess would be into the electrons. so you get faster, possibly more thermalized electrons on the outside.

just an idea.

but in any case that still would not result in a non-thermal distribution at the core, because these ions aren't all "annealing" at the same radius. the ones further out while have more distance and electric potential to accellerate through on the way to the center that the ones not as far out. and so you're going to get a distribution of ion speeds in the center that reflects that. it may not precisely by a "thermal" (guassian) distribution, but certainly it's not going to be a bunch of ions with the exact same radial momentum. (that would suggest that they all had 0 radial momentum at the same distance from the center, and we know that's not true.) and the simulation seems to atest to that.

that's why i think that if you can get some good control over what radius ionization occurs at you can get a much slimmer velocity profile for the ions at the center and thus much better fusion rates per unit energy. the best i can think of is ECR through tuned microwaves, but that's going to follow the mag fields and thus not be exactly spherical, and as i understand it it's not totally precise, it's more like a bell curve.

[KitemanSA]

As I understand it, and I may not, but as I understand it, the whole point about annealing is that particles thermalize BETTER when they are cold. They have a larger collisional cross-section And when they thermalize, they actually become more UNIFORM in energy than they start out with. The concept is that a lower rate of thermalization at high temperature, which SPREADS the energies gets overwhelmed by the higher rate of thermalization at the low temperatures which even out the energies.

At least that is how I understand it!

[93143]

fusion products presumably wouldn't have such a large preference for the cusps (or am i wrong here?) so you can use an altogether separate set of plates for those.

Dr. Nebel has pointed out that the alphas have a gyroradius of a few cm at expected reactor field strengths. They do come out the cusps. (Then they probably start to fan out, due to space charge in addition to lateral thermal spread - according to my BoE calculations, space charge spreading of the alpha beams may actually become an issue in multi-GW machines... hopefully the cusps act like magnetic nozzles and we don't get them slinging alphas all over the place with a significant fraction of their energy being lateral...)

[TallDave]

They do come out the cusps.

A good thing too -- we looked at some BOE calcs on alpha sputtering in the wiki discussion back in the pre-TalkPolywell days and it looked bad enough Art was thinking it was a showstopper.

Chacon's paper (L. Chacon, G. H. Miley, D. C. Barnes, D. A. Knoll, Phys. Plasmas 7, 4547 (2000)) says partially relaxed distributions could still get good Q values.

[happyjack27]

hopefully the cusps act like magnetic nozzles and we don't get them slinging alphas all over the place with a significant fraction of their energy being lateral...)

well my sims seem to suggest that the magnetic nozzling is pretty good. also a positively charged magrid tends to squeeze them into a stream through the center. but the plasma density in my sims is pretty low and the time simulated is probably not large enough for a thorough thermalization, so that statement is not without its caveats.

[D Tibbets]

As I understand it, and I may not, but as I understand it, the whole point about annealing is that particles thermalize BETTER when they are cold. They have a larger collisional cross-section And when they thermalize, they actually become more UNIFORM in energy than they start out with. The concept is that a lower rate of thermalization at high temperature, which SPREADS the energies gets overwhelmed by the higher rate of thermalization at the low temperatures which even out the energies.

At least that is how I understand it!

Kiteman's answer to chrismb's criticism covers the idea, but I cannot resist expanding on it. It was Chrismb who first gave me the appreciation of how much the Coulomb collision crossection increases with colder temperatures. And as K pointed out, this claimed edge annealing is indeed a thermalization process, not some strange process that violates physics. Using some assumed numbers to illustrate the point: At the core mostly radial ions might have a KE of ~ 50 KeV. As they travel outward they give up this energy to the potential well, until near the edge (some defined zone around the top of the potential well) they might have a KE of ~ 0 eV (0 +/- 100 eV) This KE spread at the edge turn around zone represents the the full Maxwell Thermalization of this population of ions. Any upscattered or downscattered ions that reach or try to pass through this zone will thermalize with other ions until they are all thermalized within the expected thermailation range for the average energy. The relatively few upscattered ions that have enough energy or chance to pick up enough energy, pass beyond this zone, entering the magnetic domain, getting turned by the magnetic field or hitting a cusp and escaping. The tremendously increased Coulomb collision crossection in this cold region dictates that this is unlikely. This means that as the ions start their next pass towards the center they start with a radial KE spread of ~ 200eV (or what ever the Maxwell thermalization would be at this average temperature of a few eV.
But the important point is that this perfectly reasonable thermalized plasma then is consistently accelerated to an additional 50,000 eV KE by the potential well. At an average temperature of a few eV, a 100 eV spread is alot. But at 50,000 eV a energy spread of +/- 100 eV it tiny (~0.2%). Unless you claim the potential well acts differently on ions that start with small initial velocity differences (that would require violation of physics) this relationship would hold.

The questions (in my mind) is how much radial thermalization occurs in one pass of the ions. So long as this is tolorable, the edge annealing would reduce (anneal) this to small values.

Transverse (angular momentum increasing) thermalization is another issue. In my mind, the acceptable behavior in this regard depends on several items. First, while tight convergence to a small core might be desired, there are limits due to vertual anode formation. Also, while it would result in a larger machine, Nebel stated that core convergence is not a critical issue.
Transverse thermalization would be slowed by the radial nature of new born/ introduced ions. As these radial ions converge towards the center the angular deflections become smaller, so that at the center only purely radial coulomb deflections are possible. This almost purely radial moving ion population would gradually relax into a transversely thermalized population. But, again, the speed of this process compared to the ion lifetime is the critical issue. I would think that the edge annealing would limit the transverse spread at the edge, but I'm uncertain. Bussard, etel did mention the disadvantage of the ions reaching high enough that they enter the magnetic domain. Because of the convex field lines, the ions (or electrons) would gain angular momentum each gyro radius loop/ deflection from the Wiffleball border.. But, due to the potential well, this is supposed to be a much less common occurrence for the ions (that plus the fact that electrons are reaching this area ~ 60 times or more frequently than the ions). I've wondered if multiple bouncing off of the magnetic surface could decrease the angular momentum (like multiple bounces off of a billiard table) , but I'm guessing the high Coulomb collisionality on the edge would dampen this effect, though it is another complication that might need to be considered in a full up simulation.

Sometimes I think speaking of Mono energetic populations gets people rightly riled, as that would be impossible. More appropriate is to speak of a narrow energy spread with some defined acceptable limit. Also, this reemphasizes the need to think of the dynamics in the machine, not just selected static situations.

Dan Tibbets